Pith. sign in

REVIEW 4 major objections 5 minor 36 references

Efficient Depth Estimation for Unstable Stereo Camera Systems on AR Glasses

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A stereo depth network for AR glasses replaces geometric rectification and the traditional cost volume with learned, hardware-friendly operations, cutting end-to-end latency by 44.5% while improving accuracy.

desk verdict Solid systems paper on replacing cost volume and rectification for stereo depth on AR glasses; the HomoDepth homography approximation is untested on near-field scenes and the D1 metric definition is inconsistent, but the core latency/accuracy claims are directionally plausible and worth refereeing. read the letter →

arxiv 2411.10013 v2 pith:JMIJNJNH submitted 2024-11-15 cs.CV cs.LG

Yongfan Liu , Hyoukjun Kwon This is my paper · ORCID
classification cs.CVcs.LG
keywords stereodepthestimationARglasseshomographycostvolumerectificationpositionalencodingmulti-tasklearninglatencyoptimizationlayernormalization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that the two most latency-heavy non-machine-learning steps in stereo depth estimation on AR glasses—image rectification and the cost volume—can be replaced by learned operators that run well on GPUs and NPUs. Its MultiHeadDepth model swaps the cost volume for a layer-normalized, group-wise dot-product block with pointwise convolution, and its HomoDepth model adds a homography-estimation head whose output becomes a 2D rectification positional encoding, so unrectified stereo pairs can go straight into the network. The reported result is that MultiHeadDepth improves accuracy by 11.8-30.3% and reduces cost-volume-plus-preprocessing latency by 22.9-25.2% relative to the ARGOS baseline, and HomoDepth cuts end-to-end latency by 44.5%. If these results hold, AR depth estimation could avoid online calibration and rectification entirely, which matters because glasses frames bend and extrinsic parameters drift during use.

What carries the argument

The carrying mechanism is the rectification positional encoding (RPE), a 2D sinusoidal positional encoding evaluated at rectified coordinates: pixels on the left use $\mathrm{PE}(q_l)$, pixels on the right use $\mathrm{PE}(H_{l\to r}q_l)$, so a world point produces similar encodings in both views and the matching layers can compare appearance and position even when the raw images are unrectified. It is enabled by a homography-estimation head sharing the encoder with the depth head. The second mechanism is the multi-head cost volume: layer normalization followed by per-disparity rolling, group-wise dot products, and pointwise convolution, which replaces cosine similarity with matrix-multiplication-friendly operators. The two are trained jointly with a homoscedastic-uncertainty loss so that homography error and depth error are weighed automatically.

What would settle it

Run HomoDepth on a fixed unrectified stereo rig while a hand or object moves from about two meters to 20 cm from the cameras, and compare AbsRel against the same model given properly rectified images: if the error gap grows sharply in the near field while far-background accuracy stays flat, the global homography assumption is the cause. A simpler version is to replace the predicted single homography with per-pixel or piecewise homographies and observe whether near-field errors fall.

Watch

Extended reading notes

Core claim

The central claim is that rectification preprocessing can be represented, not executed: instead of resampling the right image to make epipolar lines horizontal, the network predicts the 3x3 homography from the shared encoder, converts it into a rectification positional encoding (RPE) $\mathrm{RPE}(q_r) := \mathrm{PE}(H_{l\to r}q_l)$, and adds that encoding inside the cost volume so matching can use positional alignment without image warping. The traditional cost volume is replaced by a multi-head formulation that layer-normalizes both feature maps, rolls one map by each disparity, computes group-wise dot products, and fuses them with a pointwise convolution. This acts as a hardware-friendly approximation of cosine similarity while adding learnable weights and multi-head perception. On the ARGOS architecture, the authors report that MultiHeadDepth delivers 11.8-30.3% accuracy improvement and 22.9-25.2% latency reduction over ARGOS, and HomoDepth accepts unrectified images directly with a 44.5% end-to-end latency reduction, with the multi-task homoscedasticity-weighted loss cutting AbsRel error by 10.0-24.3% on misaligned stereo inputs.

Load-bearing premise

The load-bearing premise is that a single $3\times 3$ homography can represent the left-right relationship of every pixel in the scene, which Eq. (3) makes exact only when $d_l/d_r\approx 1$, i.e., for points on the central plane or far away; close objects such as hands or held items in AR scenes break this condition and would receive a systematically wrong rectification encoding.

Editorial extensions

If this is right

  • Any stereo depth model built on a cost volume can adopt the multi-head cost volume without changing its encoder-decoder, and should inherit most of the reported latency reduction on matrix-multiplication hardware.
  • AR glasses whose frames bend or whose extrinsics drift no longer need an explicit online rectification step; HomoDepth consumes the raw unrectified pair directly, removing the 15-23% rectification failure rate reported for the fast online rectification baseline.
  • End-to-end one-shot depth estimation can fit inside the 100 ms real-time budget on laptop-class GPU hardware (84.5 ms for HomoDepth versus 109.0 ms for rectified ARGOS in the paper's measurements).
  • Because the homography head shares the encoder, continuous operation can reuse a stable homography across nearby frames and run only the depth path, cutting per-frame cost further.
  • Quantization to INT8 preserves the accuracy and latency advantages, suggesting the operators survive post-training quantization needed for on-device deployment.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editor's inference: the single-homography assumption should be stress-tested on near-field content; hands, held objects, and close furniture violate $d_l/d_r\approx 1$, so the RPE would encode wrong geometry exactly where AR occlusion and manipulation matter.
  • Editor's inference: the group-pointwise cost volume is not specific to depth; optical flow, stereo matching for robotics, and multi-view stereo could swap the same operator in wherever cosine-similarity cost volumes dominate latency.
  • Editor's inference: because RPE is generated from a homography at inference time, the architecture could be extended to predict a piecewise or depth-dependent transformation (e.g., multiple homographies for foreground/background) rather than one global planar map, and the paper's robustness curves suggest the depth head tolerates small homography noise but not large errors.
  • Editor's inference: a direct comparison against classical rectification plus matching on a close-range AR interaction benchmark (hands, held objects within arm's reach) would be the decisive test of whether preprocessing can be fully eliminated in practice.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper addresses latency bottlenecks in stereo depth estimation for AR glasses. It proposes MultiHeadDepth, which replaces the conventional cost volume with a multi-head group-pointwise-convolution operator plus a LayerNorm-dot-product approximation of cosine similarity, and HomoDepth, which adds a homography estimation head and a rectification positional encoding (RPE) so that unrectified stereo images can be processed without explicit rectification preprocessing. The models are evaluated on SceneFlow, ADT, DTU, and Middlebury against ARGOS, Dynamic-Stereo, MobileStereoNet, and Selective-Stereo, with latency measurements on a laptop CPU/GPU, Jetson Orin Nano, and Snapdragon. The authors report 11.8-30.3% accuracy improvements and 22.9-25.2% latency reduction for MultiHeadDepth over ARGOS, and a 44.5% end-to-end latency reduction for HomoDepth.

Significance. If the reported results are reproducible, the paper makes a useful engineering contribution: the cost-volume replacement is hardware-friendly and general, and the homography-plus-RPE formulation is a plausible way to avoid expensive rectification on unstable camera mounts. Strengths include the public code release, evaluation on a realistic AR dataset (ADT), latency measurements on three platforms including edge-scale hardware, and an explicit robustness analysis of homography noise in Appendix E. The evaluation is empirically self-contained and I do not see a circularity problem in the DTU homography ground truth, which is derived from extrinsic parameters rather than from the depth output. However, the headline claims are stronger than the evidence: the single-homography assumption is untested for near-field AR content, the D1 metric is incorrectly defined, and no variance or statistical analysis is provided. These issues need to be addressed before the specific quantitative claims can be accepted.

major comments (4)
  1. [4.1, 4.3, 5.1, Table 4, Appendix E] The central novelty of HomoDepth rests on a planar-scene assumption that is not tested. Section 4.1, Eq. (3) derives a single homography H_{l->r} under the approximation d_l/d_r ≈ 1, which holds only for points on the central plane of the cameras or far from the cameras. Section 4.3 then applies RPE(q_r) = PE(H_{l->r} q_l) to every pixel of the right feature map. For near-field objects such as hands, held items, and nearby furniture, a single 3x3 homography cannot represent depth-dependent parallax, so the RPE encodes a systematically wrong positional relationship for those pixels, and because the RPE is added inside the matching cost, the error can actively harm the learned similarity rather than acting as a benign input warp. The evaluation does not probe this regime: ADT scenarios with people are excluded (Section 5.1), DTU contains a single object at moderate range, Sceneflow-persp is a global perspective transform, and Appendix E studies homography noise but not scene-depth composition. The claim that HomoDepth can directly process unrectified images should either be tested on close-range content or explicitly scoped to scenes where the planar approximation holds.
  2. [Appendix D, Eq. (14), Tables 2 and 4] The D1 metric is defined as an inlier rate, D1 = (1/N) sum 1(|y_hat - y| / y <= 5%), which is higher-is-better, but the paper consistently treats D1 as lower-is-better and uses it in the claimed accuracy improvements. If D1 is meant to be a bad-pixel ratio, the inequality should be reversed or the complement taken. As written, the D1 columns in Tables 2 and 4 are internally inconsistent with the text: for example, on SceneFlow the MultiHeadDepth entry (0.35) is larger than the ARGOS entry (0.23), which contradicts the reported improvement if D1 is an error rate, and contradicts the stated 'lower is better' convention if D1 is an inlier rate. This is not a cosmetic issue because D1 is part of the quantitative accuracy claims.
  3. [5.2, 5.3, Tables 2-5] All accuracy and latency numbers are reported as single point estimates with no error bars, no multiple seeds, and no repeated hardware measurements. Section 5.2 says the authors 'select the best epoch,' and if this selection is done on the test set it can inflate apparent gains; even if done on validation, the absence of seed variance makes the smaller differences hard to interpret (e.g., AbsRel 0.102 vs 0.091 on SceneFlow in Table 2). Latency measurements on hardware also vary across runs and should be reported with means and ranges. I request variance over at least three training seeds and multiple latency trials so that the claimed 11.8-30.3% accuracy improvements can be assessed.
  4. [5.6, Table 5] The edge-device latency table does not support the end-to-end claim on those platforms. Table 5 reports model-only latencies for ARGOS and notes that preprocessing delay is excluded, while HomoDepth's advertised advantage is end-to-end latency. On Snapdragon CPU, HomoDepth is actually slower than ARGOS in the reported numbers (1512 ms vs 1424 ms), and on Jetson Orin Nano CPU it is slower than MultiHeadDepth (6611 ms vs 6183 ms). Without also measuring the preprocessing cost that HomoDepth removes on the same platforms, the general claim that HomoDepth reduces end-to-end latency is only demonstrated for the laptop configuration in Table 4. Please report end-to-end measurements on the edge devices, or clearly scope the latency claim to configurations where preprocessing costs were included.
minor comments (5)
  1. [4.3] The section title contains a typo: 'Postional' should be 'Positional.'
  2. [Eq. (4)] The exponents in the positional encoding formula are ambiguous in the typeset text; they should appear as f^{i/d} and f^{(i-1)/d} to match the standard sinusoidal encoding.
  3. [Algorithm 1] The loop 'for h in 0 : head num' should iterate over 0 to head_num-1; as written, h = head_num would index outside the channel slices.
  4. [Fig. 3 caption] The caption reads 'whose size is 1 x number of channels number of heads'; it should say 'number of channels divided by number of heads.'
  5. [Abstract and Section 5.4] The claimed latency reduction ranges (22.9-25.2% and 44.5%) should specify the exact comparison basis: model-only latency, model plus rectification, or full end-to-end including calibration. The current text uses different bases in different sections, which makes the headline numbers hard to reproduce from Tables 4 and 5.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported gains are supported by external benchmarks and extrinsic-derived homography ground truth, not by fitted parameters or self-citation chains.

full rationale

The central derivations are self-contained and empirically validated. Eq. (3) derives qr = (dl/dr)Hl->r ql from standard projection equations; the dl/dr ≈ 1 approximation is an explicit modeling assumption, not a restatement of the result. The homography head is trained against ground-truth homographies computed from DTU extrinsic parameters, and depth accuracy is measured on SceneFlow, Middlebury, ADT, and DTU depth maps; neither target quantity is used to define or fit the other. The cost-volume replacement (LND and multi-head cost volume) is justified by complexity analysis and by empirical similarity-map comparison in Fig. 2, and its accuracy is evaluated on held-out datasets. The sole self-citation ([18], an XR benchmark paper by the second author) is used only to motivate the importance of depth estimation for AR/VR and is not load-bearing. The robustness study in Appendix E injects synthetic noise into the predicted homography and measures its effect on depth loss, which is a sensitivity analysis rather than a circular validation. No equation reduces the claimed prediction to its inputs by construction, and no argument rests on an unverified self-citation chain.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

No new physical entities are postulated. The paper introduces hand-selected constants w and f, an unreported head count, and learned uncertainty weights. Its geometric foundation depends on the standard pinhole model plus the approximate homography assumption for non-planar scenes, which is the main fragility for AR near-field interaction.

free parameters (4)
  • homography loss weight w = 50
    Chosen by hand in Eq. (8) to make small homography elements perceptible in the loss; no sensitivity analysis is reported.
  • RPE frequency f = 200
    Default frequency in Eq. (4), chosen to satisfy 2*pi*f > sequence length; no sensitivity analysis is reported.
  • multi-head cost volume head count = not reported
    Algorithm 1 takes head_num as an input and it controls the group-wise channel slicing, but the paper never states the value used in the experiments.
  • multi-task uncertainty weights sigma_H, sigma_D = learned from training data
    Eq. (9) treats sigma_H and sigma_D as trainable scalars that balance the homography and depth losses; they are fitted rather than derived.
assumptions (3)
  • standard math Standard pinhole projection ql = (1/dl) Kl Ml Q and qr = (1/dr) Kr Mr Q, with known intrinsics and extrinsics for the relation between image points and world points.
    Eq. (2) in Section 4.1 assumes the pinhole model with intrinsics K and extrinsics M, which is standard but not explicitly justified.
  • domain assumption For most imaged points, dl/dr is approximately 1, so a single homography can represent the left-right positional relationship of the whole stereo image pair.
    Section 4.1 introduces this after Eq. (3). It is load-bearing for HomoDepth and RPE, and it fails for close objects in near-field AR use.
  • ad hoc to paper A CNN homography head that shares the depth encoder can estimate a homography accurate enough for downstream depth estimation.
    Sections 4.2 and 4.5 rely on this design assumption. Support is limited to DTU-based experiments and a Gaussian noise sensitivity study, not real glasses bending or temporal drift.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient Depth Estimation for Unstable Stereo Camera Systems on AR Glasses." pith.science (2026). https://pith.science/paper/JMIJNJNH

@misc{pith2026241110013,
  author       = {Pith},
  title        = {Pith review of: Efficient Depth Estimation for Unstable Stereo Camera Systems on AR Glasses},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JMIJNJNH}},
  note         = {Machine review of arXiv:2411.10013}
}
read the original abstract

Stereo depth estimation is a fundamental component in augmented reality (AR), which requires low latency for real-time processing. However, preprocessing such as rectification and non-ML computations such as cost volume require significant amount of latency exceeding that of an ML model itself, which hinders the real-time processing required by AR. Therefore, we develop alternative approaches to the rectification and cost volume that consider ML acceleration (GPU and NPUs) in recent hardware. For pre-processing, we eliminate it by introducing homography matrix prediction network with a rectification positional encoding (RPE), which delivers both low latency and robustness to unrectified images. For cost volume, we replace it with a group-pointwise convolution-based operator and approximation of cosine similarity based on layernorm and dot product. Based on our approaches, we develop MultiHeadDepth (replacing cost volume) and HomoDepth (MultiHeadDepth + removing pre-processing) models. MultiHeadDepth provides 11.8-30.3% improvements in accuracy and 22.9-25.2% reduction in latency compared to a state-of-the-art depth estimation model for AR glasses from industry. HomoDepth, which can directly process unrectified images, reduces the end-to-end latency by 44.5%. We also introduce a multi-task learning method to handle misaligned stereo inputs on HomoDepth, which reduces the AbsRel error by 10.0-24.3%. The overall results demonstrate the efficacy of our approaches, which not only reduce the inference latency but also improve the model performance. Our code is available at https://github.com/UCI-ISA-Lab/MultiHeadDepth-HomoDepth

Figures

Figures reproduced from arXiv: 2411.10013 by the authors.

Figure 1
Figure 1. The latency breakdown analysis of a SOTA model ARGOS [31] and ours, MULTIHEADDEPTH & HOMODEPTH on Intel i7-12700H laptop CPU. The ”RPE” refers to the 2D recti￾fication position encoding process. ”Others” refers to all the other parts of the neural network excluding cost volume blocks, such as Conv, Norm, FC, and ReLU6. Since our methodologies are complementary to existing depth estimation models, we augment our meth… view at source ↗
Figure 2
Figure 2. Similarity estimation methodology Comparison. Each represents a similarity map between left and right features after applying roll operation (offset: 10) to the input stereo im￾ages. The maps are average results across the entire dataset of Sceneflow [21], after rescaling to [0,1] range. The strips on the left side of maps are caused by roll and they are part of the maps. ment [11]. Kendall et al. [14] adopted deep … view at source ↗
Figure 3
Figure 3. The comparison between the original cost volume and [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The structure of MULTIHEADDEPTH. The dashed lines indicate that the input activations from the left image are passed to the decoders. The input example is from ADT dataset. Algorithm 1 Multi-head Cost Volume Input: left : [NCHW], right : [NCHW], max disparity : int, he…
Figure 5
Figure 5. Figure 5: Structure of HOMODEPTH. The input example is from DTU dataset. neural network (CNN) for homography estimation. Fol￾lowing the intuition, we also designed a CNN for estimat￾ing homography matrix. Unlike the previous work, which designed a stand-alone network dedicated f…
Figure 6
Figure 6. Figure 6: Illustration of 2D rectification position encoding. Color patterns indicate the values of positional encodings. Because both PE(ql ) and RPE(qr) use the same coordi￾nates, the same location in the real world leads to similar position encoding values in stereo images. W…
Figure 7
Figure 7. Figure 7: A comparison of output qualities across models and datasets [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: The overall comparisons of models regarding # pa￾rameters, FLOPs, and latencies versus accuracy. For all co￾ordinates, the closer to the bottom left, the better model perfor￾mance. Here CPU indicates Core™ i7-12700H and GPU indicates Nvidia® GeForce RTX 3070 Ti laptop …
Figure 9
Figure 9. Figure 9: The robustness analyze of HOMODEPTH (1) How sensitive is the depth estimation to the homography estimation? (2) How stable is the homography estimation? Homography Estimation Errors. We analyze the error of HOMODEPTH during homography estimation. In prac￾tice, the erro…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 31 canonical work pages

  1. [1]

    https: / / developer

    Jetson orin nano developer kit getting started guide. https: / / developer . nvidia . com / embedded / learn / get- started- jetson- orin- nano- devkit. Ac- cessed on Nov 12, 2024. 8

    work page 2024

  2. [2]

    https://www

    Snapdragon 8+ gen 1 mobile platform. https://www. qualcomm.com/products/mobile/snapdragon/ smartphones/snapdragon- 8- series- mobile- platforms. Accessed on Nov 12, 2024. 8

    work page 2024

  3. [3]

    Jong-gil Ahn, Euijai Ahn, Seulki Min, Hyeonah Choi, Howon Kim, and Gerard J. Kim. Size perception of aug- mented objects by different ar displays. InHCI International 2019 - Posters, pages 337–344, Cham, 2019. Springer Inter- national Publishing. 1

    work page 2019

  4. [4]

    Document of torch.nn.LayerNorm

    PyTorch Contributors. Document of torch.nn.LayerNorm . Accessed on Mar 09, 2025. 11

    work page 2025

  5. [5]

    Deep image homography estimation, 2016

    Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabi- novich. Deep image homography estimation, 2016. 4

    work page 2016

  6. [6]

    Project aria: A new tool for egocentric multi-modal ai research

    Jakob Engel, Kiran Somasundaram, Michael Goesele, Albert Sun, Alexander Gamino, Andrew Turner, Arjang Talattof, Arnie Yuan, Bilal Souti, Brighid Meredith, et al. Project aria: A new tool for egocentric multi-modal ai research. arXiv preprint arXiv:2308.13561, 2023. 2, 7

    arXiv 2023

  7. [7]

    Three-dimensional computer vision: a ge- ometric viewpoint

    Olivier Faugeras. Three-dimensional computer vision: a ge- ometric viewpoint. MIT press, 1993. 2

    work page 1993

  8. [8]

    Light-weight network for real-time adap- tive stereo depth estimation

    Wanshui Gan, Pak Kin Wong, Guokuan Yu, Rongchen Zhao, and Chi Man V ong. Light-weight network for real-time adap- tive stereo depth estimation. Neurocomputing, 441:118–127,

    work page

Show all 36 references
  1. [9]

    Fast r-cnn

    Ross Girshick. Fast r-cnn. In 2015 IEEE International Conference on Computer Vision (ICCV), pages 1440–1448,

    2015

  2. [10]

    Multiple View Ge- ometry in Computer Vision

    Richard Hartley and Andrew Zisserman. Multiple View Ge- ometry in Computer Vision . Cambridge University Press, New York, NY , USA, 2 edition, 2003. 4

    2003

  3. [11]

    Evaluation of cost functions for stereo matching

    Heiko Hirschmuller and Daniel Scharstein. Evaluation of cost functions for stereo matching. In 2007 IEEE Confer- ence on Computer Vision and Pattern Recognition, pages 1– 8, 2007. 3

    2007

  4. [12]

    Large scale multi-view stereopsis eval- uation

    Rasmus Jensen, Anders Dahl, George V ogiatzis, Engil Tola, and Henrik Aanæs. Large scale multi-view stereopsis eval- uation. In 2014 IEEE Conference on Computer Vision and Pattern Recognition, pages 406–413. IEEE, 2014. 6

    2014

  5. [13]

    Dy- namicstereo: Consistent dynamic depth from stereo videos

    Nikita Karaev, Ignacio Rocco, Benjamin Graham, Natalia Neverova, Andrea Vedaldi, and Christian Rupprecht. Dy- namicstereo: Consistent dynamic depth from stereo videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13229–13239, 2023. 2, 7

    2023

  6. [14]

    End-to-end learning of geometry and context for deep stereo regression

    Alex Kendall, Hayk Martirosyan, Saumitro Dasgupta, Peter Henry, Ryan Kennedy, Abraham Bachrach, and Adam Bry. End-to-end learning of geometry and context for deep stereo regression. In Proceedings of the IEEE International Con- ference on Computer Vision (ICCV), 2017. 3

    2017

  7. [15]

    Multi-task learning using uncertainty to weigh losses for scene geome- try and semantics

    Alex Kendall, Yarin Gal, and Roberto Cipolla. Multi-task learning using uncertainty to weigh losses for scene geome- try and semantics. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7482–7491,

  8. [16]

    One shot 3d photography

    Johannes Kopf, Kevin Matzen, Suhib Alsisan, Ocean Quigley, Francis Ge, Yangming Chong, Josh Patterson, Jan- Michael Frahm, Shu Wu, Matthew Yu, et al. One shot 3d photography. ACM Transactions on Graphics (TOG), 39(4): 76–1, 2020. 1, 2

    2020

  9. [17]

    Stereo rectification of uncalibrated and heterogeneous images

    Sanjeev Kumar, Christian Micheloni, Claudio Piciarelli, and Gian Luca Foresti. Stereo rectification of uncalibrated and heterogeneous images. Pattern Recognit. Lett. , 31:1445– 1452, 2010. 1

    2010

  10. [18]

    Xrbench: An ex- tended reality (xr) machine learning benchmark suite for the metaverse

    Hyoukjun Kwon, Krishnakumar Nair, Jamin Seo, Jason Yik, Debabrata Mohapatra, Dongyuan Zhan, Jinook Song, Peter Capak, Peizhao Zhang, Peter Vajda, et al. Xrbench: An ex- tended reality (xr) machine learning benchmark suite for the metaverse. Proceedings of Machine Learning and ...

    2023

  11. [19]

    Robotic oc- clusion reasoning for efficient object existence prediction

    Mengdi Li, Cornelius Weber, Matthias Kerzel, Jae Hee Lee, Zheni Zeng, Zhiyuan Liu, and Stefan Wermter. Robotic oc- clusion reasoning for efficient object existence prediction. In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 2686–2692. ...

    2021

  12. [20]

    Scott Murdison, and Kevin W

    Hope Lutwak, T. Scott Murdison, and Kevin W. Rio. User Self-Motion Modulates the Perceptibility of Jitter for World- locked Objects in Augmented Reality . In 2023 IEEE Inter- national Symposium on Mixed and Augmented Reality (IS- MAR), pages 346–355, 2023. 1

    2023

  13. [21]

    Mayer, E

    N. Mayer, E. Ilg, P. H ¨ausser, P. Fischer, D. Cremers, A. Dosovitskiy, and T. Brox. A large dataset to train con- volutional networks for disparity, optical flow, and scene flow estimation. In IEEE International Conference on Computer Vision and Pattern Recognition (CVPR) , 2...

    2016 arXiv

  14. [22]

    Aria digital twin: A new benchmark dataset for egocentric 3d machine percep- tion

    Xiaqing Pan, Nicholas Charron, Yongqian Yang, Scott Pe- ters, Thomas Whelan, Chen Kong, Omkar Parkhi, Richard Newcombe, and Yuheng (Carl) Ren. Aria digital twin: A new benchmark dataset for egocentric 3d machine percep- tion. In Proceedings of the IEEE/CVF International Confer...

    2023

  15. [23]

    PyTorch: an imper- ative style, high-performance deep learning library

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas K ¨opf, Edward Yang, Zach DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu F...

    2019

  16. [24]

    Qualcomm Technologies

    Inc. Qualcomm Technologies. Snapdragon Neural Process- ing Engine (SNPE) Developer Guide, 2024. 8

    2024

  17. [25]

    Mobilenetv2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zh- moginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 4510–4520, 2018. 4

    2018

  18. [26]

    High-resolution stereo datasets with subpixel-accurate ground truth

    Daniel Scharstein, Heiko Hirschm ¨uller, York Kitajima, Greg Krathwohl, Nera Ne ˇsi´c, Xi Wang, and Porter West- ling. High-resolution stereo datasets with subpixel-accurate ground truth. In Pattern Recognition, pages 31–42, Cham,

  19. [27]

    Mobilestereonet: Towards lightweight deep net- works for stereo matching

    Faranak Shamsafar, Samuel Woerz, Rafia Rahim, and An- dreas Zell. Mobilestereonet: Towards lightweight deep net- works for stereo matching. In Proceedings of the ieee/cvf winter conference on applications of computer vision, pages 2417–2426, 2022. 2, 7

    2022

  20. [28]

    Neural network quantization with ai model efficiency toolkit (aimet)

    Sangeetha Siddegowda, Marios Fournarakis, Markus Nagel, Tijmen Blankevoort, Chirag Patel, and Abhijit Khobare. Neural network quantization with ai model efficiency toolkit (aimet). arXiv preprint arXiv:2201.08442, 2022. 8

    2022 arXiv

  21. [29]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017. 3, 5

    2017

  22. [30]

    A study of the scale-invariant feature transform on a parallel pipeline

    Phaneendra Vinukonda. A study of the scale-invariant feature transform on a parallel pipeline. Master’s thesis, Louisiana State University, 2011. LSU Master’s Theses,

    2011

  23. [31]

    A practical stereo depth system for smart glasses

    Jialiang Wang, Daniel Scharstein, Akash Bapat, Kevin Blackburn-Matzen, Matthew Yu, Jonathan Lehman, Suhib Alsisan, Yanghan Wang, Sam Tsai, Jan-Michael Frahm, et al. A practical stereo depth system for smart glasses. In Proceedings of the IEEE/CVF Conference on Computer Vi- sio...

    2023

  24. [32]

    Selective-stereo: Adaptive frequency information selection for stereo matching

    Xianqi Wang, Gangwei Xu, Hao Jia, and Xin Yang. Selective-stereo: Adaptive frequency information selection for stereo matching. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 19701–19710, 2024. 7

    2024

  25. [33]

    Mvsnet: Depth inference for unstructured multi-view stereo

    Yao Yao, Zixin Luo, Shiwei Li, Tian Fang, and Long Quan. Mvsnet: Depth inference for unstructured multi-view stereo. In Proceedings of the European conference on computer vi- sion (ECCV), pages 767–783, 2018. 4, 11

    2018

  26. [34]

    Mobidepth: real-time depth estimation using on-device dual cameras

    Jinrui Zhang, Huan Yang, Ju Ren, Deyu Zhang, Bangwen He, Ting Cao, Yuanchun Li, Yaoxue Zhang, and Yunxin Liu. Mobidepth: real-time depth estimation using on-device dual cameras. In Proceedings of the 28th Annual Interna- tional Conference on Mobile Computing And Networking , p...

    2022

  27. [36]

    There is no sig- nificant speed up from smartphone to laptop, showing the limitations of keypoint matching

    and our experiments, keypoint matching takes around 300ms on both smartphones and laptops. There is no sig- nificant speed up from smartphone to laptop, showing the limitations of keypoint matching. C. Dataset Setting DTU setting: Based on previous implementations and common p...

  28. [2014]

    Springer International Publishing. 7

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.